<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Cognitive Diversity on Aneesh Sathe</title><link>https://aneeshsathe.com/tags/cognitive-diversity/</link><description>Recent content in Cognitive Diversity on Aneesh Sathe</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Fri, 10 Apr 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://aneeshsathe.com/tags/cognitive-diversity/index.xml" rel="self" type="application/rss+xml"/><item><title>The Cosmos and the Model</title><link>https://aneeshsathe.com/the-cosmos-and-the-model/</link><pubDate>Fri, 10 Apr 2026 00:00:00 +0000</pubDate><guid>https://aneeshsathe.com/the-cosmos-and-the-model/</guid><description>&lt;h2 id="i-the-averaging-machine"&gt;
I. The Averaging Machine
&lt;a href="#i-the-averaging-machine" class="heading-anchor"&gt;#&lt;/a&gt;
&lt;/h2&gt;
&lt;p&gt;In 2025, researchers at USC ran a study that produced a result nobody expected.&lt;label for="marginnote-1" class="margin-toggle marginnote-ind"&gt;&lt;/label&gt;
&lt;input type="checkbox" id="marginnote-1" class="margin-toggle"/&gt;
&lt;span class="marginnote"&gt;
Atari et al., &amp;ldquo;AI-Powered Homogenization of Scientific Reasoning,&amp;rdquo; &lt;em&gt;Trends in Cognitive Sciences&lt;/em&gt; (2026).
&lt;/span&gt;
 They gave people reasoning tasks, some with LLM assistance, some without. Individual performance improved. Every metric showed it: more accurate responses, faster completion, fewer errors. The LLM made each person better.&lt;/p&gt;
&lt;p&gt;Then they looked at the groups.&lt;/p&gt;</description></item></channel></rss>